Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. A non-transitory machine-readable medium including instructions that, when executed by a machine, cause the machine to perform operations comprising: determining, based on elevation data of a geographic region corresponding to a location at which an image was captured and a solar elevation angle at a time the image was captured, whether each pixel of the image is a shadow or a non-shadow to create a shadow mask of the image; generating an eroded shadow mask that includes the shadow mask within a specified number of pixel values from a perimeter of each shadow in the shadow mask changed to respective values corresponding to non-shadows; generating a dilated shadow mask that includes the specified number of pixel values in the shadow mask changed to values corresponding to shadows; and refining the shadow mask using the image, the eroded shadow mask, and the dilated shadow mask including identifying a pixel in the dilated shadow mask that includes a value different from a corresponding pixel in the eroded shadow mask, determine, for each identified pixel, a mean shadow value based on pixels in a window of pixels centered on the identified pixel that include a shadow mask value less than a shadow threshold, determine, for each identified pixel, a mean non-shadow value based on pixels in the window that include a shadow mask value greater than a non-shadow threshold, and update the value of the shadow mask pixel based on the intensity of that pixel and the determined mean shadow value and mean non-shadow value.
This invention relates to image processing techniques for accurately identifying and refining shadow regions in captured images using elevation data and solar angle information. The problem addressed is the difficulty in distinguishing shadows from non-shadow areas in images, which can affect subsequent image analysis tasks such as object detection, mapping, or autonomous navigation. The system processes an image by first determining whether each pixel is a shadow or non-shadow based on elevation data of the geographic region where the image was taken and the solar elevation angle at the capture time. This creates a shadow mask. The mask is then refined through erosion and dilation processes. Erosion removes shadow pixels near the perimeter of detected shadows, while dilation expands the shadow regions by a specified number of pixels. These operations help adjust the shadow boundaries. Further refinement involves analyzing pixels in the dilated mask that differ from the eroded mask. For each such pixel, the system calculates a mean shadow value and a mean non-shadow value within a surrounding window of pixels. The pixel's value in the shadow mask is then updated based on its intensity relative to these mean values, improving the accuracy of shadow detection. This method enhances the precision of shadow identification, which is critical for applications requiring reliable environmental analysis.
2. The non-transitory machine-readable medium of claim 1 , wherein the operations comprise: generating a shadow histogram of non-shadow values in the complement of the dilated shadow mask; and generating a non-shadow histogram of shadow values in the eroded shadow mask.
This invention relates to image processing techniques for analyzing shadows in digital images. The problem addressed is the need to accurately distinguish between shadow and non-shadow regions in an image to improve tasks such as object detection, segmentation, or lighting correction. The solution involves generating histograms that separately represent shadow and non-shadow regions using morphological operations on a shadow mask. The method begins by processing a shadow mask, which is a binary image indicating shadow regions. The shadow mask is dilated to expand the shadow regions, and the complement of this dilated mask is used to identify non-shadow regions. A shadow histogram is generated from the original image, but only for pixels within the eroded shadow mask, which contracts the shadow regions to exclude boundary ambiguities. Similarly, a non-shadow histogram is created from the original image, but only for pixels in the complement of the dilated shadow mask, ensuring non-shadow regions are clearly defined. These histograms provide statistical distributions of pixel values for shadow and non-shadow areas, enabling more accurate analysis of lighting conditions and object properties in the image. The technique improves the robustness of computer vision algorithms by reducing errors caused by shadow misclassification.
3. The non-transitory machine-readable medium of claim 2 , wherein the operations further comprise: smoothing the shadow histogram and the non-shadow histogram; and estimating a probability that a pixel corresponds to a shadow based on the smoothed shadow histogram and the smoothed non-shadow histogram.
This invention relates to image processing techniques for shadow detection in digital images. The problem addressed is accurately distinguishing shadow regions from non-shadow regions in images, which is challenging due to variations in lighting conditions and surface properties. The invention provides a method for improving shadow detection by analyzing pixel intensity distributions and refining the results through statistical smoothing. The technique involves generating a shadow histogram and a non-shadow histogram from an input image, where each histogram represents the distribution of pixel intensities for shadow and non-shadow regions, respectively. These histograms are then smoothed to reduce noise and improve reliability. The smoothed histograms are used to estimate the probability that a given pixel corresponds to a shadow, enhancing the accuracy of shadow detection. The method may also include preprocessing steps, such as converting the image to grayscale and applying a threshold to separate potential shadow and non-shadow pixels before histogram generation. The smoothing process ensures that the histograms are more robust to outliers, leading to more precise shadow probability estimates. This approach improves the reliability of shadow detection in various applications, including computer vision, robotics, and image enhancement.
4. The non-transitory machine-readable medium of claim 3 , wherein the operations further comprise: weighting, based on a likelihood determined based on a number of shadow values in each bin of the shadow histogram and a sum of entries in a corresponding bin of the non-shadow histogram and the shadow histogram, the shadow value; and determining, based on the weighted shadow value, a weighted non-shadow value.
This invention relates to data processing techniques for analyzing histograms, particularly in systems where data may include both shadow and non-shadow values. The problem addressed is the accurate differentiation and weighting of these values to improve data analysis, such as in machine learning or statistical modeling. The invention involves a method for processing histograms where data entries are categorized into bins. A shadow histogram and a non-shadow histogram are generated, where shadow values represent data points that may be less reliable or require special handling. The method calculates a likelihood for each bin based on the number of shadow values in the shadow histogram and the sum of entries in the corresponding bins of both histograms. This likelihood is used to weight the shadow values, and a weighted non-shadow value is then determined based on the weighted shadow value. The weighting process ensures that shadow values are appropriately factored into the analysis, improving the accuracy of subsequent data processing steps. This technique is useful in applications where data quality varies, such as in sensor data analysis, anomaly detection, or noisy dataset processing. The method enhances the reliability of statistical models by accounting for the uncertainty or reliability of certain data points.
5. The non-transitory machine-readable medium of claim 4 , wherein weighting the values of the total shadow value include identifying how many pixels in the non-shadow histogram are in a bin corresponding to the value of that pixel and dividing by a total number of pixels in the bin of the non-shadow histogram and a corresponding bin the shadow histogram.
This invention relates to image processing techniques for analyzing shadows in images, particularly for distinguishing between shadowed and non-shadowed regions. The problem addressed is accurately determining the contribution of shadows to pixel values in an image, which is crucial for applications like object detection, scene reconstruction, and computer vision tasks where shadows can obscure or distort features. The invention involves a method for weighting pixel values in a shadow histogram based on their occurrence in a non-shadow histogram. Specifically, it calculates a weight for each pixel value by determining how many pixels in the non-shadow histogram fall into the same bin as that pixel value. This count is then divided by the total number of pixels in the corresponding bin of both the non-shadow and shadow histograms. This weighting process helps normalize the influence of shadowed pixels, improving the accuracy of shadow detection and analysis. The technique leverages histogram-based analysis to compare pixel distributions between shadowed and non-shadowed regions, ensuring that the shadow histogram is adjusted proportionally to the frequency of pixel values in the non-shadowed reference. This approach enhances the robustness of shadow detection algorithms by accounting for variations in lighting conditions and surface properties. The method is implemented using a non-transitory machine-readable medium, such as software or firmware, to process digital images efficiently.
6. The non-transitory machine-readable medium of claim 4 , wherein weighting the values of the total non-shadow value include identifying how many pixels in the shadow histogram are in a bin corresponding to the value of that pixel and dividing by a total number of pixels in the bin of the non-shadow histogram and a corresponding bin the shadow histogram.
This invention relates to image processing techniques for analyzing and correcting shadows in digital images. The problem addressed is the accurate determination of shadow regions in an image to improve image quality, particularly in applications like computer vision, surveillance, and medical imaging where shadow detection is critical. The invention involves a method for processing image data stored on a non-transitory machine-readable medium. The method includes generating a shadow histogram and a non-shadow histogram from the image data, where each histogram represents the distribution of pixel values in shadow and non-shadow regions, respectively. The method then weights the values of the total non-shadow value by determining how many pixels in the shadow histogram fall into a specific bin corresponding to a given pixel value. This count is divided by the total number of pixels in the corresponding bin of the non-shadow histogram and the shadow histogram. This weighting process helps distinguish between shadow and non-shadow regions more accurately by comparing the relative pixel distributions in both histograms. The technique improves shadow detection by leveraging statistical analysis of pixel value distributions, allowing for more precise segmentation of shadowed areas in an image. This can enhance subsequent image processing tasks such as object recognition, background subtraction, and illumination correction. The method is particularly useful in automated systems where accurate shadow detection is essential for reliable performance.
7. The non-transitory machine-readable medium of claim 1 , wherein the operations further include: projecting the image data to a solar coordinate system in which a direction of solar rays from the sun are perpendicular to columns of pixels of the image data; and projecting the elevation data to the solar coordinate system.
This invention relates to image processing techniques for analyzing solar energy systems, particularly for evaluating the performance of solar panels. The problem addressed is the difficulty in accurately assessing solar panel efficiency due to variations in sunlight angles and terrain elevation, which can distort measurements when using conventional imaging systems. The invention involves a method for processing image data and elevation data of a solar energy system. The image data, captured by an imaging device, is projected into a solar coordinate system where the direction of solar rays is aligned perpendicular to the columns of pixels in the image. This alignment ensures that the sunlight direction is consistent across the image, eliminating distortions caused by varying angles of incidence. Additionally, elevation data, which represents the terrain or panel orientation, is also projected into the same solar coordinate system. By aligning both datasets in this manner, the system can accurately correlate the image data with the physical layout of the solar panels, accounting for elevation differences that might affect sunlight exposure. This approach improves the accuracy of solar panel performance analysis by standardizing the imaging data relative to solar radiation, allowing for more precise detection of defects, efficiency variations, or misalignments in the solar array. The method is particularly useful for large-scale solar farms where terrain variations and panel orientations can significantly impact energy output.
8. The non-transitory machine-readable medium of claim 7 , wherein determining whether each pixel of the image is a shadow or a non-shadow to create a shadow mask of the image occurs in the solar coordinate system and the operations further include projecting the shadow mask to an image coordinate system accounting for an orientation of a camera relative to the geographic location.
This invention relates to image processing techniques for shadow detection and mapping in solar energy applications. The technology addresses the challenge of accurately identifying and analyzing shadows in images captured by cameras, particularly for solar panel installations or other solar-related systems. The method involves determining whether each pixel in an image represents a shadow or non-shadow region, generating a shadow mask that distinguishes these areas. A key aspect is performing this analysis in a solar coordinate system, which aligns with solar positioning data, and then projecting the resulting shadow mask into an image coordinate system. This projection accounts for the camera's orientation relative to the geographic location where the image was captured, ensuring accurate spatial alignment between the detected shadows and the physical environment. The technique enables precise shadow mapping for applications such as solar panel performance monitoring, energy yield optimization, or environmental analysis. By converting between coordinate systems, the method ensures that shadow data remains accurate regardless of camera placement or angle, improving reliability in solar energy assessments.
9. A method for generating a refined shadow mask, the method comprising: determining, based on elevation data of a geographic region corresponding to a location at which an image was captured and a solar elevation angle at a time the image was captured, whether each pixel of the image is a shadow or a non-shadow to create a shadow mask of the image; generating an eroded shadow mask that includes the shadow mask with a specified number of pixels from a perimeter of each shadow in the shadow mask changed to respective values corresponding to non-shadows; generating a dilated shadow mask that includes the specified number of pixels in the shadow mask changed to values corresponding to shadows; and refining the shadow mask using the eroded shadow mask and the dilated shadow mask to create a refined shadow mask including identifying a pixel in the dilated shadow mask that includes a value different from a corresponding pixel in the eroded shadow mask, detei mine, for each identified pixel, a mean shadow value based on pixels in a window of pixels centered on the identified pixel that include a shadow mask value less than a shadow threshold, determine, for each identified pixel, a mean non-shadow value based on pixels in the window that include a shadow mask value greater than a non-shadow threshold and update the value of the shadow mask pixel based on the intensity of that pixel and the determined mean shadow value and mean non-shadow value.
This invention relates to refining shadow masks in images, particularly for geographic or aerial imagery where shadows can obscure features. The method addresses the challenge of accurately distinguishing shadows from non-shadow regions in images, which is critical for applications like remote sensing, autonomous navigation, and geographic mapping. Shadows can distort the appearance of objects, making it difficult to analyze or interpret the imagery accurately. The method begins by analyzing elevation data of a geographic region and the solar elevation angle at the time the image was captured to determine whether each pixel in the image is a shadow or non-shadow, generating an initial shadow mask. To refine this mask, the method creates an eroded shadow mask by converting a specified number of perimeter pixels of each shadow region to non-shadow values, and a dilated shadow mask by converting the same number of non-shadow pixels to shadow values. The refinement process then compares the dilated and eroded masks to identify pixels where the values differ. For each such pixel, the method calculates a mean shadow value and a mean non-shadow value within a surrounding window of pixels. The pixel's value in the shadow mask is then updated based on its intensity relative to these mean values, improving the accuracy of shadow detection. This approach enhances the precision of shadow delineation, reducing errors in subsequent image analysis.
10. The method of claim 9 , further comprising: generating a shadow histogram of non-shadow values in the dilated shadow mask; and generating a non-shadow histogram of shadow values in the eroded shadow mask.
This invention relates to image processing techniques for analyzing shadow and non-shadow regions in digital images. The method addresses the challenge of accurately distinguishing between shadowed and non-shadowed areas in an image, which is critical for applications such as object detection, scene understanding, and computer vision tasks. Shadows can obscure important features and mislead algorithms, so precise segmentation is essential. The method involves generating a dilated shadow mask to expand the boundaries of detected shadow regions, ensuring that adjacent non-shadow pixels are included. This helps in capturing the full extent of shadows, which may otherwise be missed due to edge effects or noise. Additionally, an eroded shadow mask is created to shrink the shadow regions, effectively isolating the core shadow areas while excluding peripheral pixels that may be ambiguous or influenced by noise. To further refine the analysis, the method generates a shadow histogram from the eroded mask, capturing the distribution of pixel values within the confirmed shadow regions. Similarly, a non-shadow histogram is derived from the dilated mask, representing the pixel values outside the shadow areas. These histograms provide statistical insights into the intensity distributions of shadowed and non-shadowed regions, enabling more accurate segmentation and classification in subsequent processing steps. The histograms can be used for thresholding, clustering, or other statistical analyses to improve the robustness of shadow detection algorithms.
11. The method of claim 10 , further comprising: weighting, based on a likelihood determined based on the first and non-shadow histogram, shadow values; and weighting, based on a likelihood determined based on the first and non-shadow histogram, non-shadow values.
This invention relates to image processing techniques for distinguishing between shadow and non-shadow regions in an image. The problem addressed is accurately identifying and separating shadowed areas from non-shadowed areas in an image, which is crucial for applications such as object detection, scene understanding, and image enhancement. The method involves generating a first histogram representing shadow values and a second histogram representing non-shadow values. These histograms are derived from image data, where shadow values correspond to regions affected by shadows and non-shadow values correspond to regions unaffected by shadows. The method further includes determining a likelihood for each pixel or region in the image based on the first and non-shadow histograms. This likelihood indicates the probability that a given pixel or region belongs to a shadow or non-shadow category. The method then weights the shadow values and non-shadow values based on the determined likelihoods. This weighting step adjusts the contribution of shadow and non-shadow regions in subsequent processing steps, improving the accuracy of shadow detection and separation. The weighted values can be used to refine image segmentation, enhance contrast, or improve object detection algorithms by reducing the impact of shadows on image analysis. The technique is particularly useful in applications where accurate shadow detection is critical, such as autonomous navigation, surveillance, and medical imaging.
12. The method of claim 11 , wherein weighting the values of the total shadow value include identifying how many pixels in the non-shadow histogram are in a bin corresponding to the value of that pixel and dividing by a total number of pixels in the bin of the non-shadow histogram and a corresponding bin the shadow histogram.
This invention relates to image processing techniques for shadow detection and analysis in digital images. The problem addressed is accurately distinguishing between shadowed and non-shadowed regions in an image to improve computer vision tasks such as object recognition, scene understanding, or autonomous navigation. The method involves analyzing pixel values in an image to determine shadow regions by comparing histograms of shadowed and non-shadowed areas. Specifically, the technique calculates a total shadow value for each pixel by weighting its value based on statistical distributions in the histograms. The weighting process includes determining how many pixels in the non-shadow histogram fall into the same bin as the current pixel's value, then dividing this count by the sum of pixels in the corresponding bins of both the non-shadow and shadow histograms. This weighted approach helps normalize the influence of pixel values across different lighting conditions, improving shadow detection accuracy. The method also involves generating separate histograms for shadowed and non-shadowed regions, where each histogram represents the distribution of pixel intensities in those regions. By comparing these histograms, the system can identify patterns that distinguish shadows from other low-intensity areas, such as dark objects or shadows cast by objects. The weighted shadow value calculation refines this comparison by accounting for the relative frequency of pixel values in both histograms, reducing false positives in shadow detection. This technique is particularly useful in applications requiring precise scene analysis, such as robotics, surveillance, or augmented reality.
13. The method of claim 11 , wherein weighting the values of the total non-shadow value include identifying how many pixels in the shadow histogram are in a bin corresponding to the value of that pixel and dividing by a total number of pixels in the bin of the non-shadow histogram and a corresponding bin of the shadow histogram.
This invention relates to image processing techniques for analyzing and correcting shadows in digital images. The problem addressed is the accurate identification and adjustment of shadow regions in images to improve visual quality or enable further analysis. The method involves generating histograms for both shadow and non-shadow regions of an image, where each histogram represents the distribution of pixel values. The technique then weights the values in the non-shadow histogram based on the relative frequency of corresponding pixel values in the shadow histogram. Specifically, for each pixel value in the non-shadow histogram, the method counts how many pixels in the shadow histogram fall into the same bin (a range of pixel values) and divides this count by the total number of pixels in the corresponding bins of both the non-shadow and shadow histograms. This weighting process helps to normalize the influence of shadow regions when processing the image, ensuring that shadowed areas do not disproportionately affect the overall image analysis or correction. The method is particularly useful in applications requiring precise shadow detection, such as surveillance, medical imaging, or autonomous vehicle systems, where accurate scene interpretation is critical.
14. The method of claim 9 , further comprising: projecting the image data to a solar coordinate system in which a direction of solar rays from the sun are perpendicular to columns of pixels of the image data; and projecting the elevation data to the solar coordinate system.
This invention relates to image processing techniques for solar energy applications, specifically for analyzing solar irradiance data. The method addresses the challenge of accurately modeling solar radiation by transforming image data and elevation data into a solar coordinate system. In this system, the direction of solar rays is aligned perpendicular to the columns of pixels in the image data, ensuring consistent and accurate solar irradiance calculations. The elevation data, which represents terrain or surface height variations, is also projected into this coordinate system to account for topographical effects on solar exposure. By aligning both datasets in the same solar coordinate system, the method improves the accuracy of solar irradiance predictions, which is critical for solar energy assessments, such as site selection, panel efficiency optimization, and energy yield forecasting. The transformation ensures that solar radiation patterns are correctly mapped to the terrain, reducing errors caused by misalignment between solar rays and surface features. This approach enhances the reliability of solar energy modeling by integrating precise spatial and solar data.
15. The method of claim 14 , wherein determining whether each pixel of the image is a shadow or a non-shadow to create a shadow mask of the image occurs in the solar coordinate system and the operations further include projecting the shadow mask to an image coordinate system accounting for an orientation of a camera relative to the geographic location.
This invention relates to shadow detection and mapping in images, particularly for applications in solar energy assessment or geographic analysis. The problem addressed is accurately identifying and mapping shadows in images to determine their impact on solar energy generation or other location-based assessments. The method involves analyzing an image to classify each pixel as either a shadow or non-shadow, creating a shadow mask that represents the shadow regions. This classification is performed in a solar coordinate system, which aligns with solar positioning data, ensuring accurate shadow detection relative to sunlight direction. The method further includes projecting this shadow mask from the solar coordinate system into an image coordinate system, accounting for the camera's orientation relative to the geographic location. This projection step ensures that the shadow mask accurately corresponds to the real-world positions of shadows in the image, allowing for precise analysis of shadow effects. The method may also involve additional steps such as capturing the image, preprocessing the image data, and using the shadow mask for further analysis, such as solar energy potential assessment or geographic mapping. The invention improves the accuracy of shadow detection by integrating solar positioning data and camera orientation, enabling more reliable shadow analysis in various applications.
16. A system comprising: processing circuitry; a memory including program instructions that, when executed the processing circuitry, configure the processing circuitry to: determine, based on elevation data of a geographic region corresponding to a location at which an image was captured and a solar elevation angle at a time the image was captured, whether each pixel of the image is a shadow or a non-shadow to create a shadow mask of the image; generate an eroded shadow mask that includes the shadow mask with a specified number of pixels from a perimeter of each shadow in the shadow mask changed to respective values corresponding to non-shadows; generate a dilated shadow mask that includes the specified number of pixels in the shadow mask changed to values corresponding to shadows; generate a shadow histogram of non-shadow values in the dilated shadow mask; generate a non-shadow histogram of shadow values in the eroded shadow mask; and refine the shadow mask using the eroded shadow mask, the dilated shadow mask, the shadow histogram, and the non-shadow histogram including identifying a pixel in the dilated shadow mask that includes a value different from a corresponding pixel in the eroded shadow mask, determine, for each identified pixel, a mean shadow value based on pixels in a window of pixels centered on the identified pixel that include a shadow mask value less than a shadow threshold, determine, for each identified pixel, a mean non-shadow value based on pixels in the window that include a shadow mask value greater than a non-shadow threshold, and update the value of the shadow mask pixel based on the intensity of that pixel and the determined mean shadow value and mean non-shadow value.
The system is designed for shadow detection and refinement in captured images, particularly in geographic regions where elevation data is available. The problem addressed is the accurate identification and processing of shadows in images, which can interfere with various applications such as remote sensing, autonomous navigation, and computer vision tasks. Shadows can distort color, texture, and other visual features, making it difficult to analyze or interpret the image data accurately. The system uses processing circuitry and a memory with program instructions to perform several key functions. First, it determines whether each pixel in an image is a shadow or non-shadow based on elevation data of the geographic region and the solar elevation angle at the time the image was captured, creating a shadow mask. The system then generates an eroded shadow mask, where pixels near the perimeter of shadows are converted to non-shadow values, and a dilated shadow mask, where pixels near the perimeter of non-shadows are converted to shadow values. A shadow histogram is generated from non-shadow values in the dilated mask, and a non-shadow histogram is generated from shadow values in the eroded mask. The shadow mask is refined by comparing the dilated and eroded masks to identify pixels with differing values. For each identified pixel, the system calculates a mean shadow value and a mean non-shadow value based on neighboring pixels within a defined window. The pixel's value in the shadow mask is then updated based on its intensity and the calculated mean values, improving the accuracy of shadow detection. This refinement process helps distinguish between true shadows and non-shadow regions, enhancing the overall quality of the shadow mask for further analysis or processing.
17. The system of claim 16 , wherein the program instructions further configure the processing circuitry to: project the image data to a solar coordinate system in which a direction of solar rays from the sun are perpendicular to columns of pixels of the image data; and project the elevation data to the solar coordinate system.
This invention relates to image processing systems for analyzing solar radiation data. The system addresses the challenge of accurately mapping solar radiation patterns onto image data, particularly for applications like solar energy assessment or environmental monitoring. The system includes processing circuitry configured to receive image data and elevation data, where the image data represents a surface and the elevation data corresponds to the surface's topography. The processing circuitry processes the image data to correct for distortions caused by the sun's position, ensuring accurate solar radiation analysis. Specifically, the system projects the image data and elevation data into a solar coordinate system where the direction of solar rays is perpendicular to the pixel columns of the image data. This alignment simplifies calculations and improves the accuracy of solar radiation modeling by accounting for the sun's angle relative to the surface. The elevation data is similarly projected into this coordinate system to ensure consistency between the image and topographical data. This approach enhances the precision of solar radiation analysis, particularly in applications requiring detailed spatial resolution, such as solar panel placement or land surface energy balance studies. The system's ability to align solar rays with pixel columns streamlines the processing workflow and reduces computational complexity while maintaining high accuracy.
Unknown
February 4, 2020
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.